Do statins increase and Mediterranean diet decrease the risk of breast cancer?

Other substances that influence both n-3/n-6 ratio and BC risk are the cholesterol-lowering statins. The effect of statins on cancer risk is a long story and still today there is no consensus [36‐38]. The controversy began in 1996 with the publication of the Cholesterol and Recurrent Events (CARE) trial [39]. It was a double-blind randomized trial comparing the effects (versus placebo) of the cholesterol-lowering pravastatin against coronary event after myocardial infarction in 3,583 men and 576 women. Twelve out of 286 women in the statin group but only one out of 290 in the placebo group had BC at follow-up [39]. After that, most statin investigators took care not to include high-risk women in their trials [37] and carefully monitored them through repeated interim analyses for early detection of inter-group difference trends in cancer incidence. To further confuse the data, many statin trials were prematurely terminated - and it is likely that not all have been published - without valid scientific justification. Clearly, cancers diagnosed during drug trials are unlikely to be ex nihilo cancers and more likely to be dormant cancers clinically exposed by the treatment being investigated. As the process requires a minimal length of exposure, premature termination is the best way of avoiding the cancer issue in relation to any investigated drug. However, this process leads to confusion and prevents clarification of whether the investigated drug may increase cancer risk in the non-selected general population in whom the drug is then prescribed without precaution by unaware physicians. Despite this, a meta-analysis of clinical trials published in 2006 found a 33% increase in BC incidence with statins compared with a placebo [40]. It is noteworthy that confidence intervals were large (from 0.79 to 2.26) in that meta-analysis. However, there was great heterogeneity between trials (drug dosage, length of exposure) and curiously only five of the 26 randomized trials reported BC data [40], suggesting a striking lack of completeness of reporting of patient-relevant clinical trial outcomes, a well-known major source of bias and a substantial threat to the validity of clinical research findings [41]. In view of the inherent limitations of randomized trials discussed above, in particular premature termination and short follow-up, data from observational studies are critical to examine the statin-BC relationship.

In general, meta-analyses of observational studies reported no association between statin use and BC incidence. However, since high cholesterol may reduce cancer risk (see below), and as patients taking statins have spent most of their lives with high cholesterol - which is thought to lower cancer risk [37] - observational epidemiology is also facing difficulty in identifying statin cancer signals. In that context, even a lack of difference in BC risk between statin users and non-users in observational studies with long follow-up may suggest that statins increase BC risk. The recent demonstration that long-term (10-year) statin use was associated with a two-fold increase in BC risk among contemporary postmenopausal women [42] confirms the previous data suggesting that statins increase BC risk [36‐40]. Regarding statin prescription and BC recurrence specifically, a Danish study suggested that one particular highly lipophilic statin (simvastatin) may be associated with a reduced risk [43]. However, as admitted by the authors, their study suffers major limitations. Briefly, the duration of exposure was short (a median of four years), the number of recurrences was small (n = 249 among statin users) and, very important, statin users and non-users were very different at baseline. This rendered adjustments for the many confounders - knowing that factors implicated in recurrence are not necessarily similar as those implicated in incidence - and between-group comparison very problematic. Still more important and admitted by the authors [43], confounding by indication likely explains their data [44] as the major indication for statin therapy is hypercholesterolemia, which is inherently associated with lower risk of BC recurrence [45].

The next question is whether there are biological explanations for the effect of statins on BC risk. First, statins interfere negatively with the metabolism of n-3 and n-6 - that is, they decrease the n-3/n-6 ratio [46‐48] - which may in turn increase BC risk [13, 14]. Second, statins lower cholesterol, and low cholesterol is often (but not always) associated with a high cancer rate [37]. Inconsistency in the cholesterol-cancer data is likely to reflect the existence of confounding factors. One of these factors could be insulin resistance or metabolic syndrome [49, 50]. The Metabolic Syndrome and Cancer Project (Me-Can) - with more than 577,000 participants and a mean follow-up of 11.7 years - reported that cholesterol is negatively associated with BC risk, and this is a critical finding [50]. Third, a substance arising from cholesterol (dendrogenin A) is a key factor in the development of human BC [51], reinforcing the theory that high cholesterol may be protective. Fourth, statins are toxic to mitochondria [52, 53], and mitochondrial dysfunction contributes to tumorigenesis and cancer progression [54, 55]. Fifth, converging evidence supports the hypothesis that statins increase insulin resistance and new-onset diabetes, possibly (but not only) through mitochondrial toxicity in the muscles and other tissues [56‐59]. This major side effect of statins was initially underestimated with regrettable consequences, some experts even stating that ‘the cardiovascular benefits of statin therapy exceed the diabetes hazard’ [60] while the trials upon which these claims were based were obviously flawed [61, 62]. By contrast, studies indicate highly significant increases of incident diabetes among statin users [63, 64], culminating in a 70% increase among postmenopausal women in the Women’s Health Initiative [65]. At the same time, it was learned that diabetes increases BC risk [66, 67] as well as the overall risk of cancers and cancer death [68]. As diabetes is also a marker of long-standing insulin resistance - with chronically high insulin levels and high fasting blood glucose - it is critical that metabolic syndromes have also been associated with BC risk [69‐72].

Recently, investigators curiously claimed that hypercholesterolemia is a risk factor for BC and that lowering circulating cholesterol levels (or interfering with its conversion to 27-hydroxycholesterol) may be a useful strategy to prevent and/or treat BC [73]. However, the effects of 27-hydroxycholesterol were tested in rather artificial cellular and animal models of BC and hypercholesterolemia [73]. Studies using more humanized models are required before these data could have any clinical impact. Finally, statins have been shown to increase the number of immune regulatory T cells, which in turn may hinder antitumor defenses and increase cancer risk [74].

Thus, statins may increase BC risk through increased insulin resistance and new-onset diabetes, decreased n-3/n-6 ratio, cholesterol lowering, mitochondrial toxicity and an immunomodulatory effect. Statin use also results in skeletal muscle toxicity and decreased physical activity [56‐58]. For decreasing BC risk, reducing insulin resistance, metabolic syndromes and diabetes risk is beneficial, as shown with the Mediterranean diet in the next section. Additionally, international guidelines [1‐3] recommend that women aim for optimal physical activity, which is known to decrease risk for both diabetes [75, 76] and BC [1‐4]. They also recommend that women should limit weight gain, especially around menopause, to reduce BC risk [1‐3]. In that context, a recent report - 27,886 adults, 10-year follow-up - of a rapid increase in body mass index (equivalent to a 3- to 5-kg weight gain) among statin users compared with non-users is of concern [77]. Whatever the causes of that weight gain, be it reduced physical activity in relation with skeletal muscle toxicity [56‐58], increased insulin resistance or increased caloric intake [77], it may contribute to the statin-induced increase in BC risk.

Regarding diabetes risk, increased fiber intake and consumption of flavonoids and n-3 are all inversely associated with diabetes risk [78‐82]. In line with the fact that diabetes increases BC risk, not surprisingly fiber intake [83‐86], flavonoids [19, 20] and n-3 [8, 10, 13, 14] are inversely associated with BC risk.

Finally, consumption of foods with a low glycemic impact - that is, foods with a low glycemic index (GI) - is associated with lower risks of diabetes [87, 88] and BC [89‐92].

The Mediterranean diet, the traditional dietary habits of people living around the Mediterranean Sea, is a well-known healthy dietary pattern [93]. A modernized version that includes traditional Mediterranean foods (for example, olive oil, non-refined wheat bread and wine) and foods not traditionally available to Mediterranean populations (for example, canola oil, margarines, low-fat dairy products) was tested in randomized trials and resulted in health benefits [93, 94]. The combination of high fiber, high n-3/n-6 ratio, high polyphenols and low-GI foods represents a healthy dietary pattern. Adoption of such a healthy diet is clearly associated with a lower BC risk [95‐99]. Among women with early-stage BC, increased adherence to a similar healthy dietary pattern was associated with decreasing risk of overall death and death from non-BC causes (p = 0.003) [100]. There was also a trend toward less BC death, the lack of statistical significance being explained by the quite small (n = 1,900) sample size and small number of BC deaths (n = 128) [100]. More specifically, increased adherence to the Mediterranean diet pattern is also clearly associated with fewer cancers [101], specifically pancreatic [102], gastric [103], colorectal [104], hepatocellular [105], prostate [106] and breast [107‐109]. This is not unexpected since the Mediterranean diet increases the n-3/n-6 ratio on the one hand [10, 93] and on the other decreases the risk of metabolic syndrome [110, 111] and diabetes [112, 113], both of which increase the risk of cancer - including BC - and cancer deaths [66‐72]. Also, phenolic components of olive oil lowered body iron stores, which in turn may lower insulin resistance and metabolic syndrome [114]. Finally, the Mediterranean diet is an effective strategy for obtaining statistically and clinically significant weight loss [115‐117], which in turn is considered a valuable strategy to reduce BC risk and improve survival after diagnosis [1‐4].

The only limitation regarding the prevention of BC through adherence to the Mediterranean diet regards alcohol consumption. Moderate wine drinking is indeed a component of the traditional Mediterranean diet [93]. However, alcohol consumption increases BC risk [118], while the specific effect of wine is still unclear. The usual estimate for postmenopausal women who consume no more than one alcoholic drink per day is a 7% to 10% risk increase in comparison with non-drinkers [1, 2]. This is small but significant. Alcohol consumption may also increase BC recurrence [119]. Women who use postmenopausal hormones should take particular care with BC risk in relation to alcohol consumption [1‐4]. In some [120, 121] but not all [122] studies, the excess BC risk with alcohol consumption is reduced by increasing the intake of folate. Accordingly, experts have stated that the Mediterranean way of drinking alcohol - regular and moderate consumption of polyphenol-rich wine mainly with folate-rich foods - does not appreciably influence the overall risk of cancer [123]. Given that moderate alcohol consumption also reduces the risk of cardiovascular disease [124], it appears that consuming approximately one alcoholic drink per day on average, including after BC diagnosis, is associated with optimal life expectancy without compromising BC-specific survival [125‐127].